In [1]:
import PIL as pillow
%matplotlib inline
import numpy as np
import matplotlib.image as mpimg
from scipy import ndimage
from PIL import Image
from PIL import ImageFilter
from matplotlib import pyplot as plt
from matplotlib.pyplot import *
import matplotlib.image as mpimg
import os
import cv2
import time
import scipy.fftpack as fp
#from IPython.core.display import display, HTML
#display(HTML("<style>.container { width:80% !important; }</style>"))
In [2]:
def fft(imagepath):
    ## Functions to go from image to frequency-image and back
    im2freq = lambda data:fp.rfft(fp.rfft(data, axis=0),axis=1)
    freq2im = lambda f:fp.irfft(fp.irfft(f, axis=1),axis=0)


    ## Read in data file and transform
    data = np.array(Image.open(imagepath))
    freq = im2freq(data)
    back = freq2im(freq)
    
    # Make sure the forward and backward transforms work!
    assert(np.allclose(data, back))
    ## Helper functions to rescale a frequency-image to [0, 255] and save
    remmax = lambda x: x/x.max()
    remmin = lambda x: x - np.amin(x, axis=(0,1), keepdims=True)
    touint8 = lambda x: (remmax(remmin(x))*(256-1e-4)).astype(int)
    return(touint8(freq))
In [3]:
def laplacian(path):

    me_non_blur_demosaic = cv2.imread(path)
    blur = cv2.GaussianBlur(me_non_blur_demosaic,(3,3),0,0, cv2.BORDER_DEFAULT)
    laplacian = cv2.Laplacian(blur, ddepth = cv2.CV_16UC3 , ksize = 3, scale = 1,delta = 0, borderType=cv2.BORDER_DEFAULT)
    (mean, stddev) = cv2.meanStdDev(laplacian)
    laplacian_operator_std_dev = stddev[0]
    variance = laplacian_operator_std_dev**2
    return(variance)
In [4]:
# Create image objects
img_blurry_me_DEMOSAIC = ("Sample_Me_Blur/deomosaicimage.bmp")
img_not_blurry_me_DEMOSAIC =("Sample_Me-Non-Blur/deomosaicimage.bmp")
img_blurry_DEMOSAIC = ("SampleBlur/deomosaicimage.bmp")
img_not_blurry_DEMOSAIC =("SampleNon-Blur/deomosaicimage.bmp")
img_blurry_RAW = ("SampleBlur/curFrame.bmp")
img_not_blurry_RAW = ("SampleNon-Blur/curFrame.bmp")
img_blurry_me_RAW = ("Sample_Me_Blur/curFrame.bmp")
img_not_blurry_me_RAW = ("Sample_Me-Non-Blur/curFrame.bmp")
In [5]:
os.remove('saint1.tiff')
os.remove('saint2.tiff')
os.remove('saint3.tiff')
os.remove('saint4.tiff')
os.remove('saint5.tiff')
os.remove('saint6.tiff')
os.remove('saint7.tiff')
os.remove('saint8.tiff')
imsave('saint1.tiff',fft(img_blurry_DEMOSAIC))
imsave('saint2.tiff',fft(img_not_blurry_DEMOSAIC))
imsave('saint3.tiff',fft(img_blurry_RAW))
imsave('saint4.tiff',fft(img_not_blurry_RAW))
imsave('saint5.tiff',fft(img_blurry_me_DEMOSAIC))
imsave('saint6.tiff',fft(img_not_blurry_me_DEMOSAIC))
imsave('saint7.tiff',fft(img_blurry_me_RAW))
imsave('saint8.tiff',fft(img_not_blurry_me_RAW))
In [6]:
# Create image objects
blurry_me_DEMOSAIC = mpimg.imread("Sample_Me_Blur/deomosaicimage.bmp")
not_blurry_me_DEMOSAIC = mpimg.imread("Sample_Me-Non-Blur/deomosaicimage.bmp")
blurry_DEMOSAIC = mpimg.imread("SampleBlur/deomosaicimage.bmp")
not_blurry_DEMOSAIC = mpimg.imread("SampleNon-Blur/deomosaicimage.bmp")
blurry_RAW = mpimg.imread("SampleBlur/curFrame.bmp")
not_blurry_RAW = mpimg.imread("SampleNon-Blur/curFrame.bmp")
blurry_me_RAW = mpimg.imread("Sample_Me_Blur/curFrame.bmp")
not_blurry_me_RAW = mpimg.imread("Sample_Me-Non-Blur/curFrame.bmp")

Input¶

In [7]:
# Display the originals
f, axarr = plt.subplots(1,2, figsize=(24,18))
axarr[0].imshow(blurry_RAW)
axarr[1].imshow(not_blurry_RAW)
f, axarr = plt.subplots(1,2, figsize=(24,18))
axarr[0].imshow(imread('saint3.tiff'))
axarr[1].imshow(imread('saint4.tiff'))
print("Laplacian Variance Difference = " + str(abs(float(laplacian(img_blurry_RAW))-float(laplacian(img_not_blurry_RAW)))))
print("   ################################################# RAW_test_Image ########################################")
print("     "+str(laplacian(img_blurry_RAW)) + "                                                  " + str(laplacian(img_not_blurry_RAW)))
print("     "+str(float(laplacian('saint3.tiff'))) + "                                                  " + str(float(laplacian('saint4.tiff'))))

Laplacian Variance Difference = 18.6891315508
   ################################################# RAW_test_Image ########################################
     [ 1554.61858963]                                                  [ 1535.92945807]
     0.0226199292323                                                  0.0303075256363
In [8]:
f, axarr = plt.subplots(1,2, figsize=(24,18))
axarr[0].imshow(blurry_me_RAW)
axarr[1].imshow(not_blurry_me_RAW)
f, axarr = plt.subplots(1,2, figsize=(24,18))
axarr[0].imshow(imread('saint7.tiff'))
axarr[1].imshow(imread('saint8.tiff'))
print("Laplacian Variance Difference = " + str(abs(float(laplacian(img_blurry_me_RAW))-float(laplacian(img_not_blurry_me_RAW)))))
print("   ################################################# RAW_Sharp_Image ########################################")
print("     "+str(laplacian(img_blurry_me_RAW)) + "                                                  " + str(laplacian(img_not_blurry_me_RAW)))
print("     "+str(float(laplacian('saint7.tiff'))) + "                                                  " + str(float(laplacian('saint8.tiff'))))

Laplacian Variance Difference = 166.681108005
   ################################################# RAW_Sharp_Image ########################################
     [ 785.84054787]                                                  [ 952.52165588]
     0.0403411050054                                                  0.0404007469095
In [9]:
f, axarr = plt.subplots(1,2, figsize=(25,20))
axarr[0].imshow(blurry_DEMOSAIC)
axarr[1].imshow(not_blurry_DEMOSAIC)
f, axarr = plt.subplots(1,2, figsize=(24,18))
axarr[0].imshow(imread('saint1.tiff'))
axarr[1].imshow(imread('saint2.tiff'))
print("Laplacian Variance Difference = " + str(abs(float(laplacian(img_blurry_DEMOSAIC))-float(laplacian(img_not_blurry_DEMOSAIC)))))
print("   ################################################# DEMOSAIC_test_Image ########################################")
print("     "+str(laplacian(img_blurry_DEMOSAIC)) + "                                                  " + str(laplacian(img_not_blurry_DEMOSAIC)))
print("     "+str(float(laplacian('saint1.tiff'))) + "                                                  " + str(float(laplacian('saint2.tiff'))))

Laplacian Variance Difference = 9.54167016088
   ################################################# DEMOSAIC_test_Image ########################################
     [ 114.39990774]                                                  [ 104.85823758]
     0.0218227207999                                                  0.026033458978
In [10]:
f, axarr = plt.subplots(1,2, figsize=(25,20))
axarr[0].imshow(blurry_me_DEMOSAIC)
axarr[1].imshow(not_blurry_me_DEMOSAIC)
f, axarr = plt.subplots(1,2, figsize=(24,18))
axarr[0].imshow(imread('saint5.tiff'))
axarr[1].imshow(imread('saint6.tiff'))
print("Laplacian Variance Difference = " + str(abs(float(laplacian(img_blurry_me_DEMOSAIC))-float(laplacian(img_not_blurry_me_DEMOSAIC)))))
print("   ################################################# DEMOSAIC_Sharp_Image ########################################")
print("     "+str(laplacian(img_blurry_me_DEMOSAIC)) + "                                                  " + str(laplacian(img_not_blurry_me_DEMOSAIC)))
print("     "+str(float(laplacian('saint5.tiff'))) + "                                                  " + str(float(laplacian('saint6.tiff'))))

Laplacian Variance Difference = 148.070435913
   ################################################# DEMOSAIC_Sharp_Image ########################################
     [ 85.40866811]                                                  [ 233.47910403]
     0.0389796929417                                                  0.0365731705202

Applying Sobel Filter¶

In [11]:
# Let's find edges

def apply_sobel_filter(input_image):
    im = np.asarray(input_image)
    im = input_image
    sx = ndimage.sobel(im, axis=0, mode='constant')
    sy = ndimage.sobel(im, axis=1, mode='constant')
    mag = np.hypot(sx, sy)  # magnitude
    mag *= 255.0 / np.max(mag)  # normalize (Q&D)
    sob = np.hypot(sx, sy, dtype=np.float32)
    return sob
In [12]:
f, axarr = plt.subplots(1,2, figsize=(24,18))
axarr[0].imshow(apply_sobel_filter(blurry_RAW))
axarr[1].imshow(apply_sobel_filter(not_blurry_RAW))
f, axarr = plt.subplots(1,2, figsize=(24,18))
axarr[0].imshow(apply_sobel_filter(blurry_me_RAW))
axarr[1].imshow(apply_sobel_filter(not_blurry_me_RAW))
f, axarr = plt.subplots(1,2, figsize=(25,20))
axarr[0].imshow(apply_sobel_filter(blurry_DEMOSAIC))
axarr[1].imshow(apply_sobel_filter(not_blurry_DEMOSAIC))
f, axarr = plt.subplots(1,2, figsize=(25,20))
axarr[0].imshow(apply_sobel_filter(blurry_me_DEMOSAIC))
axarr[1].imshow(apply_sobel_filter(not_blurry_me_DEMOSAIC))
Out[12]:
<matplotlib.image.AxesImage at 0x1a14e87490>
In [13]:
    #font                   = cv2.FONT_HERSHEY_SIMPLEX
    #bottomLeftCornerOfText = (10,100)
    #fontScale              = 2
    #fontColor              = (255,0,0)
    #lineType               = 2
    #cv2.putText(me_non_blur_demosaic,str(float(laplacian_operator_std_dev)), bottomLeftCornerOfText, font, fontScale,fontColor,lineType)

    #Display the image
    #f, axarr = plt.subplots(1,2, figsize=(25,20))
    #axarr[0].imshow(mpimg.imread(me_non_blur_demosaic))
In [14]:
#cv2.imshow(apply_sobel_filter(img_blurry_DEMOSAIC))
In [15]:
#Different Sobel fucntions (yield slight=ly a different visual)
def pply_sobel_filter(input_image):
    im = np.asarray(input_image)
    im = input_image
    sx = ndimage.sobel(im, axis=1, mode='constant')
    sy = ndimage.sobel(im, axis=0, mode='constant')
    mag = np.hypot(sx, sy)  # magnitude
    mag *= 255.0 / np.max(mag)  # normalize (Q&D)
    sob = np.hypot(sx, sy, dtype=np.float32)
    imsave('1.tiff',sob)
    return sob


def aapply_sobel_filter(input_image):

    import numpy as np
    from scipy.signal import convolve2d
    from scipy.ndimage import imread

    img = np.asarray(input_image)
    img = input_image
    # Prepare the kernels
    a1 = np.matrix([1, 2, 1])
    a2 = np.matrix([-1, 0, 1])
    Kx = a1.T * a2
    Ky = a2.T * a1

    # Apply the Sobel operator
    Gx = convolve2d(img, Kx, "same", "symm")
    Gy = convolve2d(img, Ky, "same", "symm")
    G = np.sqrt(Gx**2 + Gy**2)
    # or using the absolute values
    #G = np.abs(Gx) + np.abs(Gy)
    return(G)

imsave('saint9.tiff',pply_sobel_filter(blurry_RAW))
imsave('saint10.tiff',pply_sobel_filter(not_blurry_RAW))

imsave('saint11.tiff',aapply_sobel_filter(blurry_RAW))
imsave('saint12.tiff',aapply_sobel_filter(not_blurry_RAW))

imsave('saint13.tiff',pply_sobel_filter(blurry_DEMOSAIC))
imsave('saint14.tiff',pply_sobel_filter(not_blurry_DEMOSAIC))

imsave('saint15.tiff',aapply_sobel_filter(blurry_DEMOSAIC))
imsave('saint16.tiff',aapply_sobel_filter(not_blurry_DEMOSAIC))

imsave('saint17.tiff',pply_sobel_filter(blurry_me_DEMOSAIC))
imsave('saint18.tiff',pply_sobel_filter(not_blurry_me_DEMOSAIC))

imsave('saint19.tiff',aapply_sobel_filter(blurry_me_DEMOSAIC))
imsave('saint20.tiff',aapply_sobel_filter(not_blurry_me_DEMOSAIC))

imsave('saint21.tiff',pply_sobel_filter(blurry_me_RAW))
imsave('saint22.tiff',pply_sobel_filter(not_blurry_me_RAW))

imsave('saint23.tiff',aapply_sobel_filter(blurry_me_RAW))
imsave('saint24.tiff',aapply_sobel_filter(not_blurry_me_RAW))
In [16]:
f, axarr = plt.subplots(1,2, figsize=(24,18))
axarr[0].imshow(aapply_sobel_filter(blurry_RAW))
axarr[1].imshow(aapply_sobel_filter(not_blurry_RAW))
print("Laplacian Variance Difference = " + str(abs(float(laplacian("saint11.tiff"))-float(laplacian("saint12.tiff")))))
print("   ################################################# RAW_test_Image ########################################")
print("     "+str(float((laplacian("saint11.tiff")))) + "                                                  " + str(float(laplacian("saint12.tiff"))))

Laplacian Variance Difference = 0.46011316237
   ################################################# RAW_test_Image ########################################
     17.590936039                                                  18.0510492013
In [17]:
f, axarr = plt.subplots(1,2, figsize=(24,18))
axarr[0].imshow(pply_sobel_filter(blurry_RAW))
axarr[1].imshow(pply_sobel_filter(not_blurry_RAW))
print("Laplacian Variance Difference = " + str(abs(float(laplacian("saint9.tiff"))-float(laplacian("saint10.tiff")))))
print("   ################################################# RAW_test_Image ########################################")
print("     "+str(float((laplacian("saint9.tiff")))) + "                                                  " + str(float(laplacian("saint10.tiff"))))

Laplacian Variance Difference = 166.981406931
   ################################################# RAW_test_Image ########################################
     1564.99384126                                                  1731.97524819
In [18]:
f, axarr = plt.subplots(1,2, figsize=(24,18))
axarr[0].imshow(aapply_sobel_filter(blurry_me_RAW))
axarr[1].imshow(aapply_sobel_filter(not_blurry_me_RAW))
print("Laplacian Variance Difference = " + str(abs(float(laplacian("saint23.tiff"))-float(laplacian("saint24.tiff")))))
print("   ################################################# RAW_Sharp_Image ########################################")
print("     "+str(float((laplacian("saint23.tiff")))) + "                                                  " + str(float(laplacian("saint24.tiff"))))

Laplacian Variance Difference = 109.484155976
   ################################################# RAW_Sharp_Image ########################################
     21.5914552654                                                  131.075611241
In [19]:
f, axarr = plt.subplots(1,2, figsize=(24,18))
axarr[0].imshow(pply_sobel_filter(blurry_me_RAW))
axarr[1].imshow(pply_sobel_filter(not_blurry_me_RAW))
print("Laplacian Variance Difference = " + str(abs(float(laplacian("saint21.tiff"))-float(laplacian("saint22.tiff")))))
print("   ################################################# RAW_Sharp_Image ########################################")
print("     "+str(float((laplacian("saint21.tiff")))) + "                                                  " + str(float(laplacian("saint22.tiff"))))

Laplacian Variance Difference = 231.013737412
   ################################################# RAW_Sharp_Image ########################################
     1498.9210148                                                  1729.93475221
In [20]:
f, axarr = plt.subplots(1,2, figsize=(25,20))
axarr[0].imshow(aapply_sobel_filter(blurry_DEMOSAIC))
axarr[1].imshow(aapply_sobel_filter(not_blurry_DEMOSAIC))
print("Laplacian Variance Difference = " + str(abs(float(laplacian("saint15.tiff"))-float(laplacian("saint16.tiff")))))
print("   ################################################# DEMOSAIC_test_Image ########################################")
print("     "+str(float((laplacian("saint15.tiff")))) + "                                                  " + str(float(laplacian("saint16.tiff"))))

Laplacian Variance Difference = 1.39218239949
   ################################################# DEMOSAIC_test_Image ########################################
     21.0678287352                                                  22.4600111347
In [21]:
f, axarr = plt.subplots(1,2, figsize=(25,20))
axarr[0].imshow(pply_sobel_filter(blurry_DEMOSAIC))
axarr[1].imshow(pply_sobel_filter(not_blurry_DEMOSAIC))
print("Laplacian Variance Difference = " + str(abs(float(laplacian("saint13.tiff"))-float(laplacian("saint14.tiff")))))
print("   ################################################# DEMOSAIC_test_Image ########################################")
print("     "+str(float((laplacian("saint13.tiff")))) + "                                                  " + str(float(laplacian("saint14.tiff"))))

Laplacian Variance Difference = 105.852760036
   ################################################# DEMOSAIC_test_Image ########################################
     1611.45459766                                                  1717.3073577
In [22]:
f, axarr = plt.subplots(1,2, figsize=(25,20))
axarr[0].imshow(aapply_sobel_filter(blurry_me_DEMOSAIC))
axarr[1].imshow(aapply_sobel_filter(not_blurry_me_DEMOSAIC))
print("Laplacian Variance Difference = " + str(abs(float(laplacian("saint19.tiff"))-float(laplacian("saint20.tiff")))))
print("   ################################################# DEMOSAIC_Sharp_Image ########################################")
print("     "+str(float((laplacian("saint19.tiff")))) + "                                                  " + str(float(laplacian("saint20.tiff"))))

Laplacian Variance Difference = 108.184136705
   ################################################# DEMOSAIC_Sharp_Image ########################################
     18.2001471542                                                  126.384283859
In [23]:
f, axarr = plt.subplots(1,2, figsize=(25,20))
axarr[0].imshow(pply_sobel_filter(blurry_me_DEMOSAIC))
axarr[1].imshow(pply_sobel_filter(not_blurry_me_DEMOSAIC))
print("Laplacian Variance Difference = " + str(abs(float(laplacian("saint17.tiff"))-float(laplacian("saint18.tiff")))))
print("   ################################################# DEMOSAIC_Sharp_Image ########################################")
print("     "+str(float((laplacian("saint17.tiff")))) + "                                                  " + str(float(laplacian("saint18.tiff"))))

Laplacian Variance Difference = 274.714591464
   ################################################# DEMOSAIC_Sharp_Image ########################################
     1418.91168479                                                  1693.62627626
In [24]:
hist1 = cv2.calcHist([apply_sobel_filter(blurry_RAW)],[0],None,[256],[0,256])
hist2 = cv2.calcHist([apply_sobel_filter(not_blurry_RAW)],[0],None,[256],[0,256])
hist3 = cv2.calcHist([apply_sobel_filter(blurry_DEMOSAIC)],[0],None,[256],[0,256])
hist4 = cv2.calcHist([apply_sobel_filter(not_blurry_DEMOSAIC)],[0],None,[256],[0,256])
hist5 = cv2.calcHist([apply_sobel_filter(blurry_me_RAW)],[0],None,[256],[0,256])
hist6 = cv2.calcHist([apply_sobel_filter(not_blurry_me_RAW)],[0],None,[256],[0,256])
hist7 = cv2.calcHist([apply_sobel_filter(blurry_me_DEMOSAIC)],[0],None,[256],[0,256])
hist8 = cv2.calcHist([apply_sobel_filter(not_blurry_me_DEMOSAIC)],[0],None,[256],[0,256])